Ammonia-gas-selective optical sensors based on ... - ACS Publications

King Tong Lau , Steve Edwards , Dermot Diamond .... Martina Trinkel , Wolfgang Trettnak , Franz Reininger , Roman Benes , Paul O'Leary , Otto S. Wolfb...
5 downloads 0 Views 1MB Size
640

Anal. Chem. 1991, 63,640-644

LITERATURE CITED

Table IV. Detection Limits of REEs (ng mL-')O ICP-MS this s t u d y element

mass no.

NEB

La Ce

139 140 141 142 152 153 158 159 164 165 168 169 174 175

0.001 0.001

Pr Nd Sm Eu Gd Tb

DY Ho Er Tm Yb Lu (I

ETV

0.0001 (2)* 0.0001 (3) 0.001 0.0001 (2) 0.003 0.0004 (8) 0.003 0.0005 (10) 0.002 0.0003 (6) 0.003 0.0006 (12) 0.001 0.0001 (2) 0.002 0.0003 (7) 0.001 0.0001 (1) 0.002 0.0003 (5) 0.001 0.0001 (2) 0.002 0.0004 (7) 0.001 0.0001 (2)

ref

7

ICP-AES ref4

NEB

NEB

0.010 0.010 0.010 0.045 0.030 0.015 0.055 0.010 0.035 0.010 0.035 0.010 0.030 0.010

4.3 31 64 22 30 0.9 18 58 25 3.6 8.0 5.1 2.2 1.o

Starred value remesents t h e absolute amount ( t ) .

heating temperature of the tungsten furnace. The memory effect due to incomplete removal of the analyte from the furnace could be reduced by increasing the heating temperature. The MO+/M+ ratios for the REEs obtained by ETV-ICPMS were 2 or 3 orders of magnitude smaller than those obtained by conventional NEB-ICP-MS, since the use of the ETV device prevented solvent (water) from entering the ICP. ETV-ICP-MS provided almost the same precision as and better sensitivity than NEB-ICP-MS. This demonstrates that ETV-ICP-MS could be an effective method for the determination of trace REEs contained as impurities in high-purity REEs and their oxides.

(1) Spedding, F. H.; Powell, J. E.; Wheelwrlght, E. J. J . Am. Chem. Soc. 1954. 76. 612. (2) Spedding, F. H.; Powell, J. E.; Wheelwright, E. J. J . Am. Chem. Soc. 1954, 76, 2557. (3) Nash, D. L. Appl. Spectrosc. 1968, 2 2 , 101. (4) Crock, J. G.; Lichte, F. E. Anal. Chem. 1982, 5 4 , 1329. (5) Palmer, K. P.; Kennedy, C.; Roser, B. Geostand. Newsl. 1983, 7 , 301. (6) Potts, P. J.; Thorpe. 0. W.; Isaacs, M. C.; Rogers, N. W. Geostand. Newsl. 1985, 9 , 173. (7) Lichte, F. E.; Meier, A. L.; Crock, J. G. Anal. Chem. 1987, 5 9 , 1150. (8) Date, A. R.; Hutchison, D. J . Anal. At. Spectrom. 1987, 2 , 269. (9) Makishima, A.; Inamoto, I.; Chiba, K. Appl. Spectrosc. 1990, 44, 91. (10) Ito, T. Shitsuryo Bunseki 1988, 3 6 , 263. (11) Douglas, D. J.; French, J. B. Spectrochim. Acta, Part B 1986, 47, 197. (12) Longerich, H. P.; Fryer, B. J.; Strong, D. F.; Kantipuly, C . J. Spectrochim. Acta. Part B 1987. 42. 75. (13) Kubota, M.; Fudagawa, N.; Kawase, A. Anal. Sci. 1989, 5 , 701. (14) Hall, G. E. M.; Pelchat. J. C.; Boomer, D. W.; Powell, M. J . Anal. At. Spectrom. 1988, 3 , 791. (15) Karanassios, V.; Horlick, G. Spectrochim. Acta, Part B 1989, 44, 1345

(16) Karanassios, V.; Horlick, G. Spectrochim . Acta, Part B 1989, 44, 1387. (17) Tsukahara. R.: Kubota, M. Spectrochim. Acta, Part B 1990, 45, 779. (18) Gregoire, D. C. J . Anal. At. Spectrom. 1988, 3 , 309. (19) Park, C. J.; Hall, G. E. M. J . Anal. At. Spectrom. 1988, 3 , 355. (20) Gregoire, D. C. Anal. Chem. 1990, 6 2 , 141. (21) Nakamura, S . ; Kobayashi, Y.; Kubota, M. Spectrochim. Acta, Part B 1986, 4 1 , 817. (22) Suzuki, M.; Ohta, K.; Yamakita, T. Anal. Chem. 1981, 5 3 , 9. (23) Tsukahara, R.; Kubota, M. Spectrochlm. Acta. Parts 1990, 4 5 , 581. (24) Park, C. J.; Van Loon, J. C.; Arrowsmith, P.; French, J. B. Can. J . Spectrosc. 1987, 3 2 , 29. (25) Mazzucotelli. A.: Frache, R . Analyst 1980, 705, 497. (26) Sen Gupta, J. G. Talanta 1961, 2 8 , 31. (27) Sincinska, P.; Michalewska, M. Fresenius' Z . Anal. Chem. 1982, 372, 530. (28) Fujino, 0.; Matsui, M. BunsekiKagaku 1982, 3 7 , 619. (29) Vaughan, M. A.; Horlick, G.; Tan, S . H. J . Anal. At. Spectrom. 1987. 2 , 765. (30) Zhu.G.; Browner, R. F. Appl. Spectrosc. 1987, 47, 349

RECEIVED for review August 20, 1990. Accepted December 13, 1990.

Ammonia-Gas-Selective Optical Sensors Based on Neutral Ionophores Satoshi Ozawa,' Peter C. Hauser, Kurt Seiler, Susie S. S. Tan, Werner E. Morf, and Wilhelm Simon* Department of Organic Chemistry, Swiss Federal Institute of Technology (ETH), Universitatstrasse 16, CH-8092Zurich, Switzerland An optical sensor (optode) that responds selectively to ammonia gas in solution samples has been realized by combining a gas-permeable membrane with a plasticized poiy(vinyi chloride) (PVC) membrane, the latter incorporating a cationselective neutral ionophore, a H+-selective neutral chromoionophore that changes its absorption spectrum upon protonation, and a lipophilic anionic site. The optodes based on NH,+-selective neutral ionophores of the macrotetrollde type showed a high selectivity for ammonia over other ammonia derivatives (e.g., log K&H3NH, = -2.7). The reproducibilities between lo-' and M ammonia samples were 4.5% and 2.6%, respectively. Optodes based on valinomycin showed moderate ammonia selectivity (log K&CH,NH, = -0.4), but the dynamic linear range was shifted to higher concentrations of ammonia so that the reproducibilities between l o 4 and M ammonia samples could be improved (3.2% and 1.6%, respectively).

' O n leave f r o m Central Research Laboratory, H i t a c h i Ltd., KO-

kubunji, T o k y o 185, Japan.

INTRODUCTION In order to meet the increasing demands for a quick and/or continuous determination of ammonia in environmental, industrial, food, biological, and clinical samples, a variety of ammonia gas sensors have been developed. These can be classified roughly into electrochemical and optical sensors. One of the former types was derived ( I ) from the Severinghaus COz electrode (2, 3) and has now become the most popular sensor for ammonia. In these sensors, a set of pH/reference electrodes is employed to detect the pH change of an internal ammonium buffer solution. This buffer solution is separated from the sample solution by a gas-permeable membrane or an air gap. By substituting an ammonium-ion-selective electrode for the pH electrode and a pH buffer solution for the ammonium buffer solution ( 4 ) ,high selectivities for ammonia over other basic gas species were obtained (5). Another kind of electrochemical gas sensor makes use of a semiconductor as the sensing device. One such sensor has been reported to possess high sensitivity and ammonia selectivity over gases such as hydrogen, methane, and acetone, but a rather high operating temperature of about 300 "C was needed (6).

0003-2700/91/0363-0640$02.50/0 0 1991 American Chemical Society

ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991

The use of optical sensors (optodes) for ammonia gas determinations has also drawn considerable attention owing to their inherent advantages such as elimination of the internal electrodes, which are expensive, bulky, and require careful handling. Since some of the ammonia optodes are not based on an internal buffer solution, they offer additional advantages of further simplification, miniaturization, and, in principle, shorter response times. An initial report on such a type of an ammonia gas optode by David e t al. focused on a ninhydrin-coated quartz rod, but the response was not reversible (7). Giuliani et al. achieved the reversibility by using a thin solid film of oxazine perchlorate as the indicator dye (8). Further investigations on this type of optode are reported by Shahriari and Zhou (9, 10). A recently published paper by Dickert et al. describes a slightly different approach that employs a triphenylmethane dye supported by a poly(viny1 chloride) (PVC) matrix (11). Another type of optode makes use of an internal buffer solution, as most of the electrochemical devices do. Arnold and Ostler reported on an optode that employs a pH indicator dye to detect the pH change of the buffer solution held behind a gas-permeable membrane (12). Wolfbeis and Posch constructed an optode on a similar principle but of a novel format; its buffer solution containing a fluorescent pH indicator was dispersed in a polymer matrix (13). In comparison t o some electrochemical ammonia sensors, selectivities for ammonia over other basic gases such as alkylamines were not attained so far with optodes. However, an optode membrane with respectable selectivities for ammonium was recently realized and described in a contribution by Seiler et al. (14). It is based on a novel principle of an optode membrane incorporating in a plasticized PVC matrix, a NH,+-selective neutral ionophore, a H+-selective neutral chromoionophore, and a lipophilic anionic site. It was mentioned that this optode membrane formally responds to ammonia, and experiments were carried out in pursuit of this possibility. By employing an additional gas-permeable membrane in order to avoid interferences from any nonvolatile ionic species in sample solutions, a new class of gas optodes with selectivities for ammonia was realized and is described here in detail.

EXPERIMENTAL SECTION Reagents. All standard solutions for the measurements were prepared with doubly quartz distilled water and salts of the highest purity available (E. Merck, Darmstadt, Germany, and Fluka AG, Buchs, Switzerland). Nonactin/monactin (about 75% nonactin, hereafter abbreviated as nonactin), dinactin, tetranactin, valinomycin, ETH 2120, bis(2-ethylhexyl)sebacate (DOS),potassium tetrakis(p-chloropheny1)borate(KTpClPB), poly(viny1chloride) (high molecular weight), and tetrahydrofuran (THF, distilled before use) were obtained from Fluka AG. Beauvericin, monensin (sodium salt), and nigericin (sodium salt) were obtained from Eli Lilly and Company (Indianapolis, IN). Bis(12-crown-4)was obtained from Dojindo Laboratories (Kumamoto, Japan). The synthesis of the H+-selectiveneutral chromoionophores ETH 2439 (1,2-benzo-7-(dimethylamino)-3[ [ p - (16-butyl-2,14-dicarbonyl3,15-dioxaeicosyl)phenyl]imino]phenoxazine),ETH 5294 (1,2benzo-7-(diethylamino)-3-(octadecanoy1imino)phenoxazine), and ETH 5350 (1,2-benzo-7-(diethylamino)-3-(2-octyldecylimino)phenoxazine) are described elsewhere (14, 15). BME-44 (2,2'tetradebis[ [3,4-(15-crown-5)-2-nitrophenyl]carbamoxymethyl] cane) (16)was kindly donated by Prof. Dr. E. Pungor (Technical University, Budapest, Hungary). Optode Preparation. Each gas optode was prepared by covering an optode membrane with a gas-permeablemembrane. The optode membranes were prepared and conditioned according to ref 14 and were washed with water after conditioning. They consisted of plasticized PVC (D0S:PVC = 2:l by weight), 27 mmol/kg of ETH 5350, 31 mmol/kg of KTpClPB, and additionally 32 mmol/kg of nonactin (in the case of composition 1) or valinomycin (composition 2). Poly(tetrafluoroethy1ene)(PTFE)

'\

4

641

SAMPLE

-

r

INCIDENT LIGHT

5 6

4

TRANSMITTED LIGHT

WASTE

Figure 1. Schematic diagram of the experimental setup. 1, sample inlet; 2, 0-seal ring; 3, glass plate; 4, gasket; 5, gas-permeable membrane; 6, optode membrane.

tape was obtained from Tecator, Hogan&, Sweden. The tape of about 20-pm thickness (17) was stretched before use to an estimated thickness of 7 pm in order to directly cover the entire surface of the optode membrane. Apparatus. Similar instrumentation as before (14) with a Uvikon Model 810 double-beam spectrophotometer (Kontron AG, Zurich) and a flow-through cell (Figure 1)was used to evaluate the optodes. The combined membranes were mounted in the flow-through cell on the detector side of the spectrophotometer. A gasket was inserted between the cell wall and the PTFE membrane to protect the fragile PTFE from puncture. No reference cell was used. Experimental Procedure. Standard solutions of the amines were prepared by adding an excess of aqueous NaOH to the respective hydrochloride salt solutions. Their pHs were above 13 so that the ammonium ions were virtually deprotonated. In addition, some ammonia-gas-buffered solutions were prepared as ammonium chloride solutions in phosphate buffer of pH 8. Their ammonia contents were calculated according to the dissociation equilibrium of ammonium ion (pK, = 9.24, T = 25 "C) by using measured pH values, calculated activity coefficients,and the total concentrations of ammonia/ammonium. Except for the full spectra measurements, absorbance measurements were made at a fixed wavelength of 640 nm. Steady-state readings for reproducibility studies were taken 10 min after the sample change. Other experimental details for optode measurements were identical with those described in the previous report (14). For potentiometric selectivity measurements, the procedures described in ref 18 were followed. Ion-selective electrodes were prepared with plasticized PVC membranes (D0S:PVC = 2:l by weight) containing 13 mmol/kg of the respective ionophore. For comparative studies, membranes additionally containing equimolar KTpClPB were also prepared. Since the mobilities and the activity coefficients of the alkylammonium ions were not available, the values of the ammonium ion were used instead.

RESULTS AND DISCUSSION The ammonium optode described in ref 14 is based on an ion exchange of H+ and NH4+between the organic membrane phase (m) and the aqueous sample (s) and therefore responds to the ratio of the respective activities or, formally, to ammonia. By separating the optode membrane from the sample solution by a gas-permeable membrane, a direct ion-exchange reaction between the sample solution and the optode membrane is inhibited. Thus, the optode membrane is left to undergo reactions with volatile and basic species (e.g., ammonia) as follows: LH+(m)

+ NH,(s) + L(m)

L(m)

+ L-NH,+(m)

(1)

where L and L stand for the neutral chromoionophore and the neutral ionophore. The electroneutrality in the optode

642

ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991

3SORBANCE

KBUFFER 10-

SAMPLE NH1 (PHOSPHATE BUFFER pH-81 A A CH~NHZ

01.

08-

10.6 M NH3 06-

f

10.5 M NH3

'

/

'10-4 M N H ~

DOS PVC

10.3 M NH3 ' 1 0 - 3 M NH3

I

I

00

NONACTIN KTpClPB DOS PVC

I

MEMBRANE ETH 5350 NONACTIN KTpClPB

10.5 M NH3

1 4

4

2

O IW [MI CONCENTRATION

500

400 Flgure 2.

A [nrr

700

600

Absorption spectra of the gas optode 1 based on nonactin for buffered solutions of ammonia.

membrane phase is ensured by lipophilic anionic sites R-, such as tetraarylborates. Since the optode membrane is based on the same formulation and the response mechanism is formally the same, essentially the same equation as that for the ammonium optode can be used to describe the theoretical response of the ammonia gas optode. The only difference is that the ratio of the activities uNHJuH has to be replaced by uNH?. Since the activity of a nonionic species does not differ largely from its concentration, aNH,can be further replaced by the concentration [NH,]. Based on a more general extension (19), the response function of the optode can be similarly derived as follows: (1 - a ) [ t T - RT

+ (1

-

a)LT]

1

-

~ [ R -T (1 - ")LT]

K,,[NH3I

(2)

where (Y is the ratio of free to total chromoionophore concentration in the optode membrane and LT, t T , and RT are the respective total concentrations of the membrane components. Ke, is the overall equilibrium constant for reaction 1 including vaporization equilibria of ammonia. The value of a is related to the absorbance A of the optode membrane by A = Ala

+ Ao(1 - a )

(3)

where A,, and A I are the respective limiting absorbance values for a = 0 and CY = 1. If interfering gases G of the concentration [GI compete with NH3 for the complexation step and if the other equilibria are not affected directly, eq 2 can be replaced by the following extended expression:

(1 - N ) [ L T - R T

+ (1 - a)LT] -

a[RT - (1 - u)LT] 1

Keq([NH31 + xK?&,GIGI)

(4)

G

where KTi3,Gare the optical selectivity coefficients (see refs 14 and 19 for details). The ammonia gas optode 1 based on nonactin showed spectral responses to buffered standard solutions of ammonia as shown in Figure 2. These spectra are comparable to those of the corresponding ammonium optode. Since the optical properties of the optode membranes were investigated in the transmission mode for these fundamental studies (see Figure l),there exists a high base-line absorbance of about 2.7 at 700

Flgure 3. Relative absorbance at 640 nm of the gas optode 1 based on nonactin as a function of the ammonia and the methylamine concentration. (0)Increasing NH, concentration, (0)decreasing NH, concentration, (A)increasing CH,NH, concentration, (A)decreasing CH3NH, concentration.

nm, which gradually increases to about 3.1 at 380 nm, due to light scattering a t the opaque P T F E membrane. In Figure 3, the relative absorbance values defined by eq 3 are plotted as a function of the concentration of ammonia and methylamine. The curve fitting the experimental points for buffered ammonia solutions was calculated from eqs 2 and 3 by using log Kq = 4.96; the curve for methylamine was calculated from eqs 3 and 4 by using the same Kq and log KTh CH3NH2 = -2.74. The experimental values for ammonia a n 8 methylamine generally fit the calculated curves well, and the response times were sufficiently fast (t9, 5 1 min) in the concentration range above 10-3,sM. However, within a limited time of measurement, the absorbance values tended to show discrepancies at concentrations lower than lo-, M, which appeared to be dependent on the direction of the change in concentrations. Since this effect is almost eliminated by introducing the same sample more than once or by using buffered solutions, it can be partly explained by the change of sample concentrations in the cell caused by an uptake into or by a discharge from the optode membrane. This problem can be taken care of by reducing the total amount of the optode membrane components, e.g., by reducing the thickness of the membrane or by decreasing the concentration of the membrane components. The resulting loss of sensitivity could be compensated by different methods of evaluation such as evanescent wave or fluorescence techniques. As described earlier, the position of the dynamic measuring range can be altered, e.g., by the use of a chromoionophore or ionophore with a different complex formation constant (for pH indicators in electrolyte solutions, see also ref 20). The initial experiments were carried out with optodes based on the chromoionophore E T H 5294, which was studied in the previous report (14). This chromoionophore and the one used here differ in their basicity; the pK, value of E T H 5350 is about 1.5 units higher than the one of ETH 5294 (15). As a result, the dynamic measuring range of the optode based on ETH 5350 was shifted to higher concentrations so that its practical working range was extended. In principle, it is possible to shift the measuring range to even higher ammonia concentrations by incorporating a more basic chromoionophore. But since ETH 5350 is currently one of the most basic chromoionophores of this type available, an ionophore having a smaller affinity for ammonium ion, while retaining the ammonia selectivity as far as possible, was searched for. Interferences from nonvolatile species are anyway excluded by the gas-permeable membrane, so only the selectivities over species

ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991

1 08

643

SAMPLE 010 N H i CHiNHp

1

AIA

I

1

MEMBRANE

4

-

rs" -E-.

-R" --Am'

I

- m a +

I

KTpClPB DOS PVC

02L

-w

I

ETH 53% VALINOMYCIN

I

1

00'

5.

I

I

6

-4

2

0

log [MI CONCENTRATION

BSORBANCE 106

1 . l A

M NH3

10-5 M NH3

Figure 6. Relative absorbance at 640 nm of the gas optode 2 based on valinomycin as a function of the ammonia and methylamine concentration. (0)Increasing NH, concentration, (0)decreasing NH, concentration, (A)increasing CH,NH, concentration, (A)decreasing CH,NH, concentration.

2 -

MEMBRANE ETH 5350 VALINOMYCIN KTpClPB DOS PVC

400

0-

500

600

700

h [nn

Figure 5. Absorption spectra of the gas optode 2 based on valinomycin for aqueous solutions of ammonia. -4

such as alkylammonium ions, which are rather volatile in basic environments, were considered. Since it was reported earlier that the selectivities of the ammonium optode are comparable with those of ion-selective electrodes (14), the search was carried out by potentiometry (see Figure 4). The ion-selective electrodes based on homologues of nonactin exhibit about the same ammonium over alkylammonium ion selectivities (Figure 4). Among the homologues, tetranactin represents the extreme, but the position of the dynamic range and the dynamic behavior of an optode membrane based on tetranactin were nearly identical with those of the optode membrane I based on nonactin. The second best potentiometric selectivity was obtained with valinomycin. Even though valinomycin has found a widespread use in ion-selective electrodes (ISEs) as an ionophore for potassium ions, gas optodes based on valinomycin showed reasonable response to ammonia gas as well. The spectra of a valinomycin-based optode membrane 2 (Figure 5) were comparable to those of nonactin-based optode membranes. The experimentally obtained relative absorbance plots showed a nearly perfect fit to the expected curves (Figure 6), which were calculated as before by using log Keq= 3.60 and log K T & C H 3 N H , = -0.38. As compared to the nonactin-based optode, the dynamic measuring range of the valinomycinbased optode for ammonia is shifted to higher concentrations (in log Keq,the difference is 1.36). Thus, in the case of the valinomycin-based optode, the deviations from the theoretically expected behavior in the lower measuring range observed

c1

- w3

- PrNH2 - EtNHZ -

Me2NH

- Me3N

nonactin valinomycin blank Flgure 7. Selectivities of ammonia gas over several alkylamine gases obtained with optodes 1 and 2 based on nonactin and valinomycin, respectively. The blank test was carried out with an optode without an ionophore and with a different chromoionophore (ETH 2439). for the nonactin-based optode have been reduced. This is mainly given by the smaller complex formation constant of valinomycin with ammonium ions as compared to nonactin. The selectivity (log K?$,,CH,NH, = -0.38) of this valinomycin-based optode is moderate as compared to the nonactinbased one, but ammonia is still the most preferred species among all the alkylamines tested (see Figure 7). An optode membrane based on the same composition as membrane 2 but without an ionophore shows a negligible response up to 0.1 M ammonia. Another optode membrane with the chromoionophore ETH 2439 and without an ionophore shows a response to ammonia at concentrations of loe3M and higher, due to its pK, being about 3.2 units lower than that of ETH 5350 (15). The selectivities of the latter (blank) optode are also shown in Figure 7. It can be readily seen that the new ionophore-based optodes offer a significant improvement in selectivity for ammonia over other basic gases. Interference from species such as NaHC03 and Na2S03, which may affect the results by generating acidic gases (COz and SO2, respectively), was also checked. If the samples were

644

ANALYTICAL CHEMISTRY, VOL. 63, NO. 6, MARCH 15, 1991

ABSORBANCE (640nml lO”MNH,

-

-

3.073

10 min

1

3.221

7

[,I A

-

3.219

3.217

A-3.p0t0.002

3.073 I _

3.221

,------

3.073

3.074

IO-~MNH, ~ - 3 , 0 7 3 i o . m i

I TIME

Flgure 8.

and

Short-time reproducibilities for sample changes between lo4 the valinomycin based optode 2.

M NH3 with

prepared with sodium hydroxide, no significant interference was observed from either of the species for both types of optode membranes (1 and 2). This is as expected because these species stay virtually in the unprotonated form under such highly basic conditions. In the case of the nonactin-based optode, the samples are preferably prepared with a buffer the pH of which is substantially lower than the pK, of the ammonium ion so that it is capable of supplying and accepting a sufficient amount of ammonia to and from the optode membrane. Under these conditions, interference from COP was particularly anticipated because the bicarbonate ion becomes partly protonated due to its relatively high pK, of 6.35 (2‘ = 25 “C). Its interference was nevertheless very small. The apparent concentration change was about -3% at the ammonia concentration of [NH,] = 6.9 X M when the carbonic acid was coexisting at a concentration of [H2C03]= 1.2 x 10-4 M. Figure 8 shows an example of short time reproducibilities of a valinomycin-based optode. In this case, the standard deviations in absorbance of 0.002 and 0.001 for and M ammonia samples correspond to 3.2% and 1.6% on the (linear) concentration scale, respectively. The same experiment with the nonactin-based optode showed similar standard deviations on the absorbance scale, but due to the smaller sensitivity in this range, reproducibilities on the concentration scale were 4.5% and 2.6%, respectively. The reason for the rather poor ammonia selectivity over methylamine of the valinomycin-based gas optode (log K$i,,CH,,NH, of about -0.4) as compared to the ammonium selectivity over methylammonium of the valinomycin based ion-selective electrode (log I@&+,CHINH;l+ of about -1.8, Figure 4) may be due to the borate anion that is present in the optode membrane. A poor selectivity was also observed for ion-selective electrodes even in the case of a nonactin-based system if an equimolar amount of borate was present in the membrane (log Kf$$+,CHINHl+ of about 0.4). It is interesting that, in the case of the nonactin-based optode, the presence of the borate does not affect its inherent high selectivity. In the case of optodes, we expect that the selectivity is induced primarily by the relative stabilities of ion-ionophore complexes, which are abundant in the membrane, and that the effect of less stable and less populated ion-borate complexes is not important, unless the stability of the ion-ionophore complex is reduced to such an extent that the influence of ion-borate complexes becomes significant. This latter situation probably holds in the case of valinomycin-based optodes. In the case of ion-selective electrodes, on the other hand, the selectivities among ions arising from different mobilities of ions induced by the ionophores can be diminished, if components such as

borates provide large mobilities without specificity to ions, to an extent that the mean mobilities of different ions become nearly the same. This may occur even if the stability and, accordingly, the population of the ion-borate complexes are comparatively small. Since a gas optode for ammonia was the main subject of this study, only ionophores having high selectivity for ammonia over alkylamines were tested. However, with the proper choice of the membrane components and its composition, it appears possible to construct a variety of gas optodes with different selectivity characteristics, suitable for diverse fields of applications.

CONCLUSIONS We have presented in this report new ammonia gas optodes that can be applied to aqueous sample solutions. The major advantage of these devices over the previous ammonia optodes is the high ammonia selectivity attained by the use of an ionophore in the recognition process. The optodes based on nonactin and its homologues showed excellent ammonia selectivities over other basic gases. Another one based on valinomycin also works as an ammonia-selective optode, which exhibits a different selectivity behavior and a dynamic range at higher ammonia concentrations. The new ammonia optodes have, together with some of the previous ones, advantages over the electrochemical ammonia gas sensors such as simplicity (especially the elimination of the internal buffer solution), possibility to determine dry gas-phase samples, small size, and suitability for qualitative detections by the naked eye. These factors may lead the optodes to completely new fields of applications so far unknown to the electrochemical gas sensors. ACKNOWLEDGMENT We thank P. Gehrig and P. Schmidli of ETH for helpful discussions. LITERATURE CITED Okada, M.; Matsushita, H. J . Chem. SOC. Jpn., Ind. Chem. Sect. 1989, 72, 1407-1409. Stow, R. W.; Baer, R. F.; Randall, 6. F. Arch. Phys. M e d . Rehabil. 1957, 38, 646-650. Severinghaus, J. W.; Bradley, A. F. J . Appl. Physiol. 1958, 73, 515-520. Meyerhoff, M. E. Anal. Chem. 1980, 52, 1532-1534. Fraticelli, Y. M.; Meyerhoff, M. E. Anal. Chem. 1981, 53, 1857-1861. Nanto, H.; Minami, T.; Takata. S. J . Appl. Phys. 1986, 60, 482-484. David, D. J.; Willson, M. C.; Ruffin, D. S. Anal. Len. 1978, 9 , 389-404. Giuliani, J. F.; Wohltjen, H.;Jarvis, N. L. Opt. Left. 1983, 8, 54-56. Shahriari, M. R.; Zhou, Q.;Sigel, G. H., Jr. Opt. Len. 1988, 73, 407-409. ZhoU, Q . ; Kritz, D.; Bonnell, L.; Sigel, G. H., Jr. Appl. Opt. 1989, 28, 2022-2025. Dickert, F. L.;Schreiner, S. K.; Mages, G. R.; Kimmel, H. Anal. Chem. 1989, 6 1 , 2306-2309. Arnold, M. A.; Ostler, T. J. Anal. Chem. 1988, 58, 1137-1140. Wolfbeis, 0.S.; Posch, H. E. Anal. Chim. Acta 1988, 785, 321-327. Seiler, K.; Morf, W. E.; Rusterholz. 6.; Simon, W. Anal. Sci. 1989, 5 , 557-561. Seiler, K. Diss. No. 9221, ETH Zurich, 1990. Lindner, E.; TBth, K.; Jeney, J.; Horvith. M.; Pungor, E.; Bitter, I.;Agai, 6.; Toke, L. Mikrochim. Acta 1990, 1 , 157-168. Schulze, G.; Brodowski, M.; Elsholz, 0.; Thiele, A. Fresenius 2.Anal. Chem. 1988, 329, 714-717. Meier, P. C.; Morf, W. E.; bubli, M. W.; Simon, W. Anal. Chim. Acta 1984. 156. 1-8. Morf, W. E.; Seiler, K.; Sorensen, P. R.; Simon, W. In Ion-Selective Nectmdes; Pungor, E., Ed.; Akademiai KiadB: Budapest, 1989; Vol. 5, pp 141-152. Rhines, T. D.;Arnold, M. A. Anal. Chem. 1988. 60, 76-81.

RECEIVED for review September 11,1990. Accepted December 13, 1990. S. Ozawa gratefully acknowledges support from Hitachi Ltd. for his stay at ETH.